Skip to main content

Full text of "Proposed characterization of tornadoes and hurricanes by area and intensity"

See other formats


V 1 















SATELLITE & MESOMETEOROLOGY 

RESEARCH PROJECT 

IhpartmmU of the Geophrsietd ^-i^m-es 



PROPOSED CHAMCTERIZATIOX OF HORKADOES AXD 
HURRICAXES BY AREA A:\D LXTFXSITY 



.-V./-1 



■i 'Tl 




-;> 



1 i • - i t r. 




T, Tlieodore Fujita 
Hie Unh'ersit}' of Chicago 



'. /,:-! 



-.-.•: '■'•> 






'! / 







SMBF MiBsmimh Paper 
Nambar 91 



Satellite and Mesometeorology Research Project 
Department of the Geophysical Sciences 
The University of Chicago 



Proposed Characterization of Tornadoes and 
Hurricanes by Area and Intensity 



Dy 



T. Theodore Fujita 
The University of C3iicago 



SMRP Research Paper No. 91 
February 1971 



The Tornado Watch Experiment in 1968 througjh 70 as well as the establishment 
of the Fujita -scale wind and damage concepts have been sponsored by NASA under grant 
NGR-14-001-008 and by NOAA under NESC grant E-198-68(G). Test analyses of 
tornado characterization were sponsored by NSSL under annual grants since 1957. 

TTie characterization of Hurricanes and Typhoons has been sponsored by NOAA 
under NHRL grant E-22-24-71(G). 



TABLE OF CONTENTS 



1. INTRODUCTION 2 

2. PROPOSED SCALE OF DAMAGING WIND 3 

Table of F-Scale Damaging Wind Speed 6 

3. SPECIFICATION OF DAMAGE CORRESPONDING TO F-SCALE WIND SPEEDS 7 

F-Scale Damage Specifications 8 

F-Scale Damage Pictures 11 

4. APPLICATION OF F SCALE IN TORNADO ANALYSES 12 

5. APPUCATION OF ESTIMATED F SCALES FOR PERIODIC 
INTENSIFICATION OF FAMILY TORNADOES 15 

6. RANGES OF INDIVIDUAL TORNADO AREA 17 

7. CHARACTERIZATION OF TORNADOES BY AREA AND INTENSITY 18 

8. APPLICATION OF F SCALE TO THE SURVEY OF HURRICANE DAMAGE . . 22 

9. USE OF TABLE IN CONVERTING MEASURED WINDS INTO F SCALE 

WINDS FOR HURRICANES 29 

Table to Determine F-Scales from Measured Wind 30 

10. CHARACTERIZATION OF HURRICANES AND TYPHOONS 35 

11. CONCLUSIONS 38 

REFERENCES 40 

SUBJECT INDEX 42 



Proposed OiaracterizatiGn of Tornadoes and 
Hurricanes by Area and titensity 



by 



T. Theodore Fujita 
The Lfaiversity of Chicago 



ABSTRACT 

Results of the 1968 through 1970 Tornado Watch Experiment conducted jointly by 
NASA and NOAA suggested the necessity of characterizing individual tornadoes in order 
to improve the identity of tornado -producing nephsystems. An attempt was made, there- 
fore, to categorize each tornado by its intensity and area. Fujita -scale wind and corre- 
sponding damage categories were devised to classify tornadoes as Gale ( FO ), Weak ( F 1 ), 
Strong ( F? ), Severe ( F 3 ), Devastating ( F 4 ), and Incredible ( F 5 ). Additionally, indi- 
vidual tomido areas were also categorized as Trace (TR), Decimicro (DM), Micro (MI), 
Meso (ME), Macro (MA), Giant (GI), and Decagiant (DG), thus permitting us to charac- 
terize a tornado by a combination of intensity snd area, such as "Weak Decimicro Tor- 
nado", "Severe Meso Tomacto", "Incredible Giant Tornado", etc. A test characteriza- 
tion of 156 Japanese tornadoes in 1950-69 was accomplished for comparison with 893 U. S. 
tornadoes in 1%5. Unexpectedly, the percentage distritxition of intensity and individual 
area of U. S. and Japanese tornadoes is very similar except for large and/or intense ones. 
Intensity distributicm within the Dallas and Fargo tornadoes of 1957 was also studied in 
detail. It was also found that the F-scale variation along the paths of family tornadoes 
shows an intensity oscillation with a 45-min interval. For further applications, charac- 
terization of Atlantic hurricanes, Pacific hurricanes, and Pacific typhoons was made to 
determine the trend of their cumulative frequencies. It was found that 90% of these 
storms are characterized, in each region respectively, Ysy less than F2. 8, F2. 3, and 
F3. 3, indicating clearly tiiat average typhoons are more intense than average hurricanes. 
Finally, the areas of Hurricane Camille of August 1969 and the Ise-wan tyi*ioon of 
September 1959 are analyzed with F-scale contours in an attempt to determine the distri- 
bution of damaging winds within these storms. 



i. INTRODUCTION 

During a three-year j)eriod, 1968-1970, annual tornado watch experiments were 
ccHiducted under the sponsoisiiip of NASA and NOAA in an attempt to investigate 
satellite -viewed cloud characteristics in relation to tornado occurrences. The evaluation 
of tornado and nejrtisystem relationship turned out to be rather inconclusive because those 
storms reported as tornadoes are not always large and intense, destroying only weak 
structures during their few-minute life time. Scwne storms are, on the ether hand, long 
lived with their incredible intensity resulting in total destruction of even steel structures. 

In an attempt to improve our basic understanding of tornado- producing thunder- 
storms viewed from geostationary satellites , past tornado data based on storm data and 
other sources were reexamined so as to classify individual siorms according to their 
intensity as well as their affected area. 

The frequency of reported tornadoes since 1916, when systematic tabulation began, 
has been increasing during the past half century although the frequencies in the 1960s 
leveled off to a certain extent. Such a trend is shown clearly in statistical papers by 
Wolford (1960), Thom (1%3), Pautz (1969), and Court (1970). Tlie importance of mean 
duration in relation to mean length of travel was emphasized by Battan (1959) who derived 
the mean duration as less than 2 minutes. For the assessment of tornado probability at 
given points in the United States, the mean area of tornadoes is of vital importance. The 
mean area was not well known until Thom (1963) estimated the mean area to be 2. 8 sq. 
miles. Such an estimate is absolutely nec^sary to compute tornado probabilities since 
the larger the mean area the higher the damage probability even if the total frequency of 
tornadoes remains unchanged. 

An attempt to categorize tornadoes, all if possible, was made by Fujita (1970) who 
used those recorded in Storm Data for 1965, 1967, and 1969 with reported lengths and 
widths. Analysis of 793 tornadoes in these three years revealed that the individual tornado 
areas obtained by simply multiplying the length by the width recorded in Storm Data vary 
through 6 orders of magnitude from 0.001 to 100 sq. miles. 

Tornado intensity is much harder to estimate than tornado affected area because 
the extent of damage to structures and trees cannot as easily be expressed numerically. 
By examining damage descriptions in Storm Data, it was felt first that a reasonable scale 
of damage can be established if an educated guess is made. The Chicago tornado of 



March 4, 1961, studied by Biown and Fujita (1961), should be rated as much weaker than 
the Dallas tornado of April 2, 1957 which was, however, considerably weaker than the 
Fargo tornado of June 20, 1957. The author made a detailed aerial survey of 25 Palm 
Sunday tornadoes of April 11, 1970 and found diat none of the 25 was as intense as the 
Fargo tornado, despite the fact that it was a day of tremendous outbreak of large and 
intense storms. When the author surveyed the Lubbock tornado r>{ May 11, 1970, 
Fargo -type, intense damage was found in many parts of the (kimage area. Thus, the 
feasibility of characterizing Chicago to LulAock type tornadoes into 3 or 4 categories 
app)ears to be realistic. By adding one or two weaker categories below the Oiicago tor- 
nado category, it would be a logical attempt to differentiate damage caused by 50 to about 
300 -mph winds into 5 to 6 categories through an educated guess by means of the charac- 
terization method described in the following paragraphs. 

2. PROPOSED SCALE OF DAMAGING WIND 

Although the frequency of tornadoes in the midwestem United States is the higjtiest 
in the world, the probability of oteerving a tornado ircm a fixed location is slightly higher 
than once in a century. A chance that a wind recorder happens to be in the immediate 
vicinity of a passing tornado is extremely rare. Moreover, most wind instruments are 
not designed to record wind speeds in excess of 150 mph and their survival in much higher 
wind speeds is unlikely. 

Wind speeds in tornadoes have been estimated from structural damage, charac- 
teristic ground marks, scanning and scaling motion pictures, and the shape of funnel 
clouds. Each of these estimates requires weeks of time in order to achieve the highest 
possible accuracy. On the other hand, one may be able to make extremely rough estimates 
of wind-speed ranges through on-the-spot inspection of storm damage. For instance, the 
patterns of dama^^e caused by 50 and 250-mph winds are so different that even a casual 
observer can recognize the differences immediately. The logic involved is that the higher 
the estimate accuracy the longer the time required to make the estimate. Thus, a few 
weeks of time necessary for an estimate with a 5 -mph accuracy can be reduced drastically 
to a few seconds if only a 100 mph accuracy is permissible in order to obtain a large 
number of estimates with considerably less accuracies. An important compranise is to 
make an educated guess of the speed ranges by inspecting either actual damage or aerial 



photographs taken after a storm. A survey of the literature reporting the results of 
time-consuming, accurate estimates implies that 30-miA increments within relatively low 
wind-speed ranges and 50-mph increments witfiin rather high wind-speed ranges result in 
characteristic damage patterns which can be distinguished by trained individuals with the 
help of damage specifications similar to those used in estimating the Beaufort wind forces 
by trained oteervers. Only a few seconds will be required to perform such an educated 
guess. An existing Beaufort wind force extended beyond B 12 is too small a speed incre- 
ment to distinguish damage corresponding to one force or another. 

The proposed scale of damaging wind is designed to connect smoothly the Beaufort 
force, B12, with the speed of sound in the atmosphere or Mach number. Ml , This scale 
which may be called F scale corresponds to Beaufort force and Mach number at the 
following wind speeds . 

Fl = Beaufort 12 
and F12=Machl at -3**C. 

Equations for canputing B, F, and M wind speeds are: 

V = 0.836B* (m/sec) = 1.870 B* (mph) (1) 

1 1 

V = 6.30 (F + 2)* (m/sec) = 14.1 (F + 2)' (mph) (2) 

V = (331+0.6t)M(m/sec) = (742 + 1.3t)M (mph) (3) 

where B, F, and M are, respectively, Beaufort force, Fujita scale, and Mach number, 
V the corresponding wind speed, and t , the air temperature in ° C. 

In deriving Eq. (2) for F scale wind computation, a linear formula such as for 
Mach number was avoided because it is desirable to have a small speed increment when 

3 

speed is relatively low. In obtaining such a speed range, (F + 2 )* was introduced. This 
simple function has three advantages: (a) the square of the speed or the energy is 
proportional to the cube of (F + 2 ), (b) the speed corresponding to F = represents a 
finite value of 40 mph which would cause little damage on most structures , and (c) the 
speed ranges between FO and Fl, Fl and F2, etc. increases 33, 40, 45, 49 mph, etc., 
resulting in distinguishable speed ranges increasing with F numbers. 



Due to the fact that F- scale winds are estimated from structural and/or tree 
damage, die estimated wind speed aj^lies to die height of the apparent damage above the 
ground. In addition to the speed and the height of damaging wind, duration is also of vital 
importance to the yielding point of a structure. While disregarding the gust period and 
amplitude at this moment, the period of sustained wind must be specified. Internationally, 
ten-minute average values are used in synoptic reports, while one-minute average values 
are used in the United States. A station equij^d with a multiple register determines the 
fastest speed in milrs per hour of any "mile" wind. At 60 mprti, the fastest-mile wind 
speed and one-minute mean speed are identical. 

In connecticm with definitions of one-minute average, 10 -minute average, and the 
fastest mile winds, it would be necessaiy to clarify the definition of die F scale wind speed 
ccMnputed from Eq. (2). A 10-min period or even a one-minute period wind seems to be 
too long to be required to destroy a structure or to blow down a tree. The period of a 
S'istained wind required to complete a damage is likely to be inversely proportional to die 
wind speed, suggesting strongly that damaging wind must be defined as the fastest "wind 
padi length" rather than die average speed during a sustained wind period. When the wind 
path length is selected to be "one mile", we call the wind the fastest mile wind. In reality, 
however, only a fraction of one mile would be required to complete a destruction of trees 
and structures. F-scale wind speed in this paper was thus defined to be the "fastest 1/4 
mile wind" . For F 4 wind speed of about ^0 mph, the duration of die damaging wind 
would be only about 4 seconds. 

Damaging wind speeds in mph, knots, and m/sec together with die pieriods of the 
fastest 1/4 mile winds are presented in Table I. Note diat Fl.O corresponds to 73 mph, 
the beginning of hurricane wind. In view of possible occurrences of lig^t damage when 
F is larger than 0.0 or 40 mph, the table was made to include FO.O and higher speeds. 
Under die presumption that die occurrences of F6 or higher wind are extremely rare, 
wind speeds above F6 are tabulated for the range of each full scale, F7, F8, etc. 
Presented also in the table are damage categories to be expected by FO throu^ F5 winds. 

In addition to the speed ranges corresponding to FO, Fl, F2, etc. , the table 
includes the fractional F scales such as Fl. 3, F2. 8, etc. These values do not imply 
that the fractional F scales can be estimated through visual inspections of damage. The 
fractional F scales can be computed only from anemometer record or engineering 



Table I. TABLE OF FUJITA SCALE DAfU^ING WIND SPEED 
F-scale wind speed is defined as the fastest ^-mile wind at the 
height of damaged structure of object. The last column indicates 
the period of the fastest %-mile wind in seconds. 









F 


( 40 


- 72 mph ) 


Light Deunage 
















F 1 


( 73 


- 112 mph ) 


Moderate Damage 














F 2 


( 113 


- 157 mph ) 


Considerable 


Damage 














F 3 


{ 158 


- 206 mph ) 


Severe Dcunage 














F 4 


( 207 


-260 mph ) 


Devastating Deunage 














F 5 


( 261 


- 318 mph ) 


Incredible 


Damage 








F-scale 


mph 


knots 


m/sec 


Period 


F 0.0 


— 


0.1 


40 - 


45 


35 


— 


39 


17.8 


— 


20.5 


22.5 


— 


19.6 


0.2 


- 


0.3 


46 - 


51 


40 


- 


45 


20.6 


- 


23.3 


19.5 


- 


17.2 


0.4 


- 


0.5 


52 - 


58 


46 


- 


50 


23.4 


- 


26.3 


17.1 


- 


15.3 


0.6 


- 


0.7 


59 - 


65 


51 


- 


56 


26.4 


- 


29.4 


15.2 


— 


13.7 


0.8 


- 


0.9 


66 - 


72 


57 


- 


63 


29.5 


- 


32.6 


13.6 


- 


12.3 


F 1.0 


— 


1.1 


73 - 


80 


64 


— 


69 


32.7 


— 


36.0 


12.2 


— 


11.2 


1.2 


- 


1.3 


81 - 


87 


70 


- 


76 


36.1 


- 


39.4 


11.1 


- 


10.2 


1.4 


- 


1.5 


88 - 


95 


77 


- 


83 


39.5 


- 


42.9 


10.1 


— 


9.4 


1.6 


- 


1.7 


96 - 


103 


84 


— 


90 


43.0 


- 


46.6 


9.3 


— 


8.7 


1.8 


- 


1.9 


104 - 


112 


91 


- 


97 


46.7 


- 


50.3 


8.6 


- 


8.1 


F 2.0 


— 


2.1 


113 - 


120 


98 


— 


104 


50.4 


— 


54.1 


8.0 


— 


7.5 


2.2 


- 


2.3 


121 - 


129 


105 


- 


112 


54.2 


- 


58.1 


7.4 


— 


7.0 


2.4 


- 


2.5 


130 - 


138 


113 


- 


120 


58.2 


- 


62.1 


6.9 


— 


6.5 


2.6 


- 


2.7 


139 - 


147 


121 


— 


128 


62.2 


- 


66.2 


6.4 


— 


6.2 


2.8 


- 


2.9 


148 - 


157 


129 


- 


136 


66.3 




70.3 


6.1 


- 


5.8 


F 3.0 


- 


3.1 


158 - 


166 


137 


— 


144 


70.4 


— 


74.6 


5.7 


— 


5.5 


3.2 


- 


3.3 


167 - 


176 


145 


- 


153 


74.7 


- 


79.0 


5.4 


- 


5.2 


3.4 


- 


3.5 


177 - 


186 


154 


- 


161 


79.1 


- 


83.4 


5.1 


— 


4.9 


3.6 


- 


3.7 


187 - 


196 


162 


- 


170 


83.5 


- 


87.9 


4.8 


- 


4.7 


3.8 


- 


3.9 


197 - 


206 


171 


- 


179 


88.0 


- 


92.5 


4.6 


- 


4.4 


F 4.0 


— 


4.1 


207 - 


217 


180 


— 


188 


92.6 


— 


97.2 


4.3 


— 


4.2 


4.2 


- 


4.3 


218 - 


227 


189 


- 


197 


97.3 


- 


101.9 


4.1 


- 


4.0 


4.4 


- 


4.5 


228 - 


238 


198 


- 


207 


102.0 


- 


106.7 


3.9 






4.6 


- 


4.7 


239 - 


249 


208 


- 


216 


106.8 


- 


111.6 


3.8 


- 


3.7 


4.8 


- 


4.9 


250 - 


260 


217 


- 


226 


111.7 


- 


116.6 


3.6 


- 


3.5 


F 5.0 


— 


5.1 


261 - 


271 


227 


_ 


235 


116.7 


_ 


121.6 


3.4 






5.2 


- 


5.3 


272 - 


283 


236 


- 


245 


121.7 


- 


126.7 


3.3 






i;.4 


- 


5.5 


284 - 


294 


246 


- 


255 


126.8 


- 


131.9 


3.2 


- 


3.1 


5.6 


- 


5.7 


295 - 


306 


256 


- 


266 


132.0 


- 


137.1 


3.0 






5.8 


- 


5.9 


307 - 


318 


267 


- 


276 


137.2 


- 


142.5 


2.9 






F 6.0 


— 


6.9 


319 - 


380 


277 


— 


329 


142.6 


— 


170.0 


2.8 


_ 


2.5 


F 7.0 


- 


7.9 


381 - 


445 


330 


- 


386 


170.1 


- 


199.1 


2.4 


- 


2.1 


F 8.0 


- 


8.9 


446 - 


513 


387 


- 


446 


199.2 


- 


229.7 


2.0 


- 


1.8 


F 9.0 


- 


9.9 


514 - 


585 


447 


- 


508 


229.8 


- 


261.8 


1.7 


- 


1.6 


FIO.O 


- 


10.9 


586 - 


660 


509 


- 


573 


261.9 


- 


295.2 


1.5 






Fll.O 


- 


11.9 


661 - 


737 


574 


- 


640 


295.3 


- 


329.9 


1.4 


- 


1.3 


F12.0 


- 


12.9 


738 - 


818 


641 


- 


710 


330.0 


- 


365.9 


1.2 


- 


1.1 



estimates of storm damage. 

Figure 1 was prepared to show the connection of Beaufort force, Fujita Scale and 
Mach number. According to the summary prepared by List (1958), Bea«ifort forces are 
numerically extended to 317 (131 mi^) which corresponds to F2. 4. By extending the 
F-scale wind speed downward below F 1. 0, it is seen that the curve ends at wind speed 
zero corresponding to F = -2. Mach number wind speed for a given air temperature is 
expressed by a straight line on which fractional Mach numbers such as 0. 6, 0. 7, ... 
are indicated. 



-»oo«,h COMBINATION OF B, F. AND M LINES 

fMIZ 
V - I 8708'^ mph - 0.8368'$' m/wc 
V- l4l(F+2)+ mph - 630(F*2)+ m/»«c 
h •« V « (742+ l3t)M nyh - (331 ♦OeDM ki/mc 



tnfM 

n»—F I20#MACH 1.0 




«00- 



FOR HURRICAME 



^■^ 



01 2345678910 II 12 
I BEAUFORT FORCE 



0I234967S9I0III2 
^^ FUJITA SCALE 



MACH NUMBER 



Fig. 1. Connection of Beaufort force, 
Fujita scale and Mach number. In 
deriving the equation for F-scale wind 
computaticHi, the following considerations 
were made. (1) To connect Beaufort 
force 12 witli Mach number 1 widi a 
smooth curve, (2) To correspond B 12 
with F I and M I with F 12, so tfiat a 1 
through 12 graduated scale, as in the case 
Beaufort force, covers die desired speed 
range. (3) Beaufort indicates calm or 
no wind and Fujita likewise denotes the 
wind speed causing no damage on most 
structures, (4) To give wider speed range 
as the speed increases because the faster 
the wind speed the wider the speed range 
to allow a visual distinction of damage 
frcon one scale tu the next, and (5) An 
exponent 3/2 is JiJcely to serve the above 
purpose. Furthermore, the square of 
the speed or the kinetic energy is propor- 
tional to the cube of F + 2. About 20 
formulas to satisfy partial or total 
conditions listed above were examined 
before adopting Eq (2), the final equation, 
which was used to oi^ain the F-scale curve 
presented in this figure. 



3. SPECIFICATION OF DAMAGE CORRESPONDING TO F-SCALE WIND SPEEDS 

F-scale wind speeds introduced in the previous section are nothing but the speeds 
computed from Eq. (2) which was formulated to connect Beaufort force with Mach number. 
It is then necessary to obtain damage specifications corresponding to each F scale. Since 
no direct measurements of wind speed inside tornadoes are available, various estimates 



of wind speeds and corresponding damage characteristics were studied in detail. 

Early estimates of the highest tornado wind speed reported by Flora (1954) and 
some others are as higji as 500 mpli. Engineering estimates revealed, however, that up 
to 350 mph winds lasting a few seconds are likely to produce most tornado damage in the 
Midwest. A comprehensive summary of tornado wind speeds based on various methods 
was completed by Melaragno (1968) in an attempt to assess tornado forces and their 
effects on buildings. 

Australian tornadoes were summarized by Clarke (1962). He stated that the 
"Brigjh ton tornado", Melbourne, February 2, 1918, which severely damaged well-con- 
structed buildings was accompanied by an estimated 200 -mph wind which was the highest 
estimated speed in Australia. Clarke indicated that 98% of Australian tornadoes are 
characterized by winds less than 120 mph and 72% by winds less than 73 mph v^ich 
corresponds to the Beaufort 12 or F 1 scale. 

Based on these estimated speeds along with a large number of aerial and ground 
photographs of tornado damage, FO throu^ F5 damage specif i^-ations were obtained. 

FUJITA SCALE DAMAGE SPECIFICATIONS 

( F ) 40-72 mph, LIGHT DAMAGE 

Some damage to chimneys And TV antennae; breaks twigs 
off trees; pushes over shallow rooted trees. 

( F 1 ) 73 - 112 mph, MODERATE DAMAGE 

Peels surface off roofs; windows broken; li^t trailer houses 
pushed or overturned; some trees uprooted or snapped; moving 
automobiles pushed off the road. 73 mph is the beginning of 
hurricane wind speed. 

( F 2 ) 113-157 mph, CONSIDERABLE DAMAGE 

Roofs torn off frame houses leaving strong upright walls; weak 
buildings in rural areas demolished; trailer houses destroyed; 
large trees snapped or uprooted; railroad boxcars pushed over; 
lig^t object missiles generated; cars blown off highway. 



8 



( F 3 ) 158-206 mph, SEVERE DAMAGE 

Roofs and some v^t'ls torn off frame houses; seme rural 
buildings completely danolished; trains overturned; steel- 
framed hangar -war^ouse type structures torn; cars lifted 
off the ground; most trees in a forest uproc^ed, snapped, or 
leveled. 

( F 4 ) 207-260 mph. DEVASTATING DAMAGE 

Whole frame houses leveled, leaving piles of debris; steel 
structures badly (^unaged; trees debarked by small flying 
debris; cars and trains thrown seme distance or rolled 
considerable distances; large missiles generated. 

( F 5 ) 261 - 318 mph. INCREDIBLE DAMAGE 

Whole frame houses tossed off foundations; steel -reinforced 
concrete structures badly damaged; automobtle-sized 
missiles generated; incredible phenomena can occur. 

( F6-12 ) 319 mph to sonic speed. INCONCEIVABLE DAMAGE 

Should a tornado with the maximum wind speed in excess 
of F 6 occur, the extent and types of damage may not be 
conceived. A number of missiles such as ice boxes , 
water heaters , storage tanks, autcHnotxles. etc. will 
create serious secondary damage oa structures. 

The above damage specifications were based mostly on engineering estimates of 
wind speeds, involving both drag and lift forces which are assumed to be proportional to 
the square of the wind speed. As shown in the schematic outline of a house in Fig. 2, a 
straight flow impinging against a house creates a positive pressure on the upwind wall A. 
Due to the acceleration of the flow around the comers , a significant ne^tive dynamic 
pressure is created at B, C, and D. Negative dynamic pressure on the roof is usually 



HURRICANE B LARGE -DIAMETER TORNADO 




FAST-MOVING. SMALL- DIAMETER TORNADO 




Fig. 2. Schematic flow 
patterns and induced dy- 
namic pressure around a 
house standing in tomsdic 
wind. Note that the re<kic- 
tion of pressure bjr a 
passing tornado as a whole 
acts as an explosive pres- 
sure superimposed upon 
the wind-induced dynamic 
pressure. 



more significant than on walls, resulting in the peeling of roofing materials or removal 
of the roof without damaging upright walls. Thus a portioQ of the debris from a house 
can be found evai on the upwind side of standing walls. If the speed of damaging wiikI 
exceeds that of F 3, walls start collapsing. With F 5 or higher wind, the drag force is 
so large that all walls of a frame house can be torn from the foundation and the lift force 
creates a flying house. 

A fast-moving tornado with a relatively small core diameter creates an additional 
pressure reduction problon. If a house is air tig^t, the hydrostatic reduction of pres- 
sure due to an approaching tornado incr^^ses the in-house pressure relative to the out- 
side pressure. This pressure which may be called the "explosive pressure' , P^ , will 
be considerably reduced when a house is properly vented, especially on the downwind 



10 




Fig, 3. F-scale damage chart applicable to relativetv new sufcrban structures. Damage scenes were 
selected from color [rtcnires of Lubtock tornado, M&y il, 1«70, taken b>' the author from about 5(X)-ft 
altitude. 



11 



side. Although the amount of P^ whicli depends upon the translational speed and the 
pressure profile of the storm as well as the venting mode of a structure, cannot be esti- 
mated for each damage case, the explosive pressure does accelerate the roof removal or 
house explosion for a given range of F-scale wind speeds. As a result, one could over- 
estimate an F scale by one. Such an overestimation can be avoided, however, by 
examining nearby structures and trees especially when an exploded house is surrounded 
by undamaged or sli^tly damaged structures. 

To aid F-scale determinations from damage specifications, the damage chart of 
Fig. 3 was prepared. This ».,iiart is applicable to most suburban structures with block 
foundations and walls and relatively weak roofing. No damage scene for FO is included 
in the chart because the damage is insignificant in aerial pictures. It is seen in the F 1 
scene that some to most roofing materials are peeled off frame houses. The F2 scene 
shows a typical aerial view of a house with its torn roof and standing upright walls. Since 
the Lubbock tornado of May 11, 1970, with an extremely large core diameter, was moving 
slowly when this damage occurred, the roof was torn mostly by wind-induced dynamic 
pressure similar to that experienced in intense hurricanes. As shown in the F 3 scene, 
some upri^t walls were torn from a motel building; individual block-sized missiles were 
found smck in ground-floor walls. The F4 scene shows a leveled structure with all items 
excel* the foundation dislocated from their original positions. Trees around this building 
were debarked by small flying debris which are usually captured by tree trunks at low wind 
speeds. The F5 scene taken in the northern suburb of Lubbock clearly shows the founda- 
tion of a house which had sailed away toward the lower rig^t, leaving behind a water 
heater and a bath tub. All trees around the house were flattened, losing most of the bark 
from their trunks. 

4. APPLICATION OF F SCALE IN TORNADO ANALYSES 

Damage charts and specifications introduced in the previous chapter can now be 
used in determining F-scale variations along the paths of well -documented storms. 

Investigations of the Dallas tornado of April 2, 1957 by Hoecker (1960a) (1960b), 
Beebe (1960), and Segner (1960) were used to produce the life history chart of Fig. 4. The 
top graph was constructed by plotting as a function of time the heists of the funnel 
appearing in Hoecker's (1960a) 57 sketches. These sketches were used to determine the 



12 







Fig. 4. Life Mstory of Dallas toimaito of April 2, 1957 as dfepicted ty the 
heigirt: and shape of the funnel, vailatioB of P-scale damage, and tihe estimated 
wind spe©ils along tbe tornadto path. 



13 



LIFE HISTORY OF FARGO TORNADO. JUNE 9a |957 




Fig. S. Life history of Fargo toniadto of Juw 20, 1957. Not€ that this stoitn 
was larger a»d more Intense than Dallas tornado in Fig, 4. 



14 



schematic variation of the funnel. Shown in the bottom chart in the figure is the tornado 
path with areas covered by aerial jrfiotos appearing in Beebe's (1960) analysis. The loca- 
tions of engineering estimates by Segner (1960) are identified by Roman numerals, I, II, 
in, etc. The middle diagram shows the variation of F-scale damage estimated by the 
author based on aerial photographs. Despite the fact that the F scale varies practically 
from house to house, one must accept such variations as the built-in noise superimposed 
upon overall patterns of tornado damage. Segner's estimated values ranging between 55 
and 189 mph (one 302 mph value which he stated doubtful) seem to fit the F-scale varia- 
tions determined from aerial photographs. These test analyses suggest that it would be 
hi^ly desirable to determine house-to-house variation of F-scale intensity in order to 
interpret properly the time-consuming engineering estimates of maximum and/or mini- 
mum wind speeds causing specific damage. 

A similar life history chart of the Fargo tornado of June 20, 1957 (see Fig. 5) was 
constructed based on the storm analysis by Fujita (1960). The Fargo tornado funnel was 
a giant compared to the Dallas tornado. When the funnel was ou the ground, the srorm 
caused F 5 damage between 25th and 29th Streets shown in the bottom chart. At that 
time, the funnel diameter on the ground was estimated to be about 500 ft. Thereafter, the 
funnel was lifted 4 to 500 ft above the ground, but the damage scale beneath the lifted 
funnel varied between F 1 and 3, suggesting that the damaging wind at the building levels 
was between 75 and 200 mjrfi. As shown in the middle chart, the rotation rate of the 
funnel shortly after its formation was 107 mprfi. By adding a 35 -mph trans lational speed 
of the tornado, the combined speed at the funnel level should be 142 mprfi. The wind speed 
affecting the building beneath the funnel is likely to be smaller than this value. 

5. APPLICATION OF ESTIMATED F SCALES FOR PERIODIC INTENSIFICATION 
OF FAMILY TORNADOES 

It has been known that tornadoes often spawn from a parent cloud in the form of 
a family. Fujita (1963) pointed out that the average occurrence interval is about 45 
minutes, Darkow and Ross (1970) studied 7 parent storms in Missouri, 1964 through 1968, 
obtaining average occurrence intervals of 45 min. As shown in Table II, the intervals are 
mostly independent of the storms traveling speed ranging between 15 and 62 mph. 



15 



Tabic II. Occurrence intervals of family tornadoes spawn from their parent storms. 
Bi-state tornado in Fig. 6 was added to the original tabulation in Fujita (1963) 
vihich was made based on analyses by Van Tassel (1955), Penn et al (1955). Staats 
and Turrentine (1956). Hoecker (1960), and Fujita (1960). 



Date of Tornado 


Tornado 




Path 


Tornado 


Occurrence 


Speed of 




Identification 


I^ength 


Duration 


Interval 


Parent Storm 


27 June 1955 


Scottsbluff 


#1 
#2 
#3 


8 

4 
9 


miles 


? 
29 
45 


min 


38 
42 


min 


15 mph 


20 June i957 


Fargo 


#1 

#2 
#3 
*4 
#5 


11 

7 

8 

10 

7 


miles 


45 
35 
63 
35 
25 


min 


40 
42 
43 

40 


min 


19 mi^ 


2 April 1957 


Dallas 


*1 


14 


miles 


35 


min 


27 


min 


26 mph 






#2 


3+ 


? 










9 June 1953 


Norcestcr 


#1 
#2 


45 
29 


miles 


75 
65 


min 


65 


min 


30 mph 


25 May 1955 


Black«Mll 


#1 

«2 


39 

37 


miles 


60 
50 


min 


40 


min 


31 mph 


11 April 1965 


Bi-state 


#1 
#2 
#3 
#4 
#5 
#6 


21 
51 
58 
38 
18 
24 


miles 


20 
49 
56 
37 
17 
23 


min 


23 
49 
64 
57 
46 


min 


62 mph 


Average of 


all storms 




22. 


1 miles 


42. 


5 min 


44. 


min 


30.5 mph 



6 FAMILY TORNADOES (INDIANA -OHIO) OF PALM SUNDAY, 1965 



50 



100 



ISO 



200 



^ 



250 274 mles 



d 



MEAN CENTER PERIOD. 48 min 



mile 
0.7- 

0.6- 

0.5- 

0.4- 

0.3- 

0.2- 

01- 

0.0- 




I 



49fflin » 



4hr23min 



in 
O 




O 

lO 
(M 
CM 



Ul 

o 

< 



F4 



M 



Z 
Ui 

z 



F3 




F2 




Fl 



J ' ' ■ I ' ' ■ ' ■ I ' ' ' 



1800 



1900 



2000 



2100 



2200 



Fig. 6. Variation of damage width and F scale along the bi-state tornado of Palm Sunday, 1965. This 
storm, leaving a straight- line damage path of 274 miles from south of Lafayette, Ind. to Cleveland, Ohio, 
clearly shows 6 maxima in both damage width and storm intensity. 



16 



The proposed F-scale intensity of tornadoes when applied to a series of family 
tornadoes will permit us to describe the variation of the intensity as a function of either 
time or space. An example of 6 family tornadoes on Palm Sunday, 1965 is shown in 
Fig. 6. Note that F-scale intensity dropped down below F 1 five times during the storms 
274 mile path in 4 hr 23 min. During this period the times when no damaging winds were 
on the ground were no longer than 30 minutes. The figure shows that the damage width or 
the diameter of damaging wind also changed periodically in phase with the storm's intensity. 
This is against die original expectation that the intensification will take place when the 
storm's diameter shrinks. Although the reasons for such an in-phase variation of the 
intensity and the damage diameter have not been solved, research is in progress to deter- 
mine the time variation of the circulation around famil^^ tornadoes expressed as 



— = — f Vt d| (4) 

dt dt Jr * ^^^ 



C 

v^ere P denotes the circulation; V^ , the tangential velocity around a closed circuit C; 
and d£ the circuit element around the storm. 

6. RANGES OF INDIVIDUAL TORNADO AREA 

For the purpose of making the maximum use of Storm Data, as was done by 
3attan (1959), Thom (1963), and odiers, both lengths and widths of individual tornadoes 
were investigated closely. The individual tornado areas defined by Fujita (1970) and 
applied to each tornado recorded in Storm Data are found to vary through an extremely 
wide range such that the area ratio between the largest and the smallest exceeds 
1,000,000 to 1. The author, therefore, tried to classify tornadoes according to their 
areas as shown in Table IE. It is very unlikely that a tornado with its individual area in 
excess of 1000 eq. miles ever occurred in the past. ITie lerm HG (hectogiant) is, there- 
fore, not included in the table. The accuracy of area estimates depends entirely upon the 
values given in Storm Data. Quoting Rattan's (1959) statement, "Interviews with witness 
and damage surveys should pay particular attention to establishing the path length and 
duration of individual tornado funnels", the author, as one of the users of Storm Data, 
also wishes to stress the importance of the original data given in Stonn Data for basic 
and applied research on tornadoes. 



17 



TR (trace) 





< a < 


0.001 


DM (decimicro) 


0.001 


<a < 


0.01 


M I (micro) 


0.01 


<a < 


0.1 


ME (meso) 


0.1 


<a < 


1 


MA (macro) 


1 


<a < 


10 


G I (gian:y 


10 


<a < 


100 


DG (decagiant) 


100 


<a < 


1000 



Table III. Classification of tornadoes according to individual 
areas defined by a = L X w where L and w are the length and 
the mean width given in Storm Data. 

Area Category a in sq. mile log a 

log a < -3 

-3 < log a < -2 

-2< log a < -1 

-1 < log a < 

< log a < +1 

1 < log Q < +2 

2 < log a < +3 

As specified in Table II, the variation of the individual tornado area from one area 
category to the next is a factor of 10, meaning that relatively inaccurate estimates will 
still allow the selection of the most reasonable category. 

7. CHARACTERIZATION OF TORNADOES BY AREA AND INTENSITY 

Estimation of both tornado intensity and area by F-scale intensity and log a scale 
area is an initial improvement over merely counting the number of tornadoes , each of 
which is identified as a unit tornado. An attempt was made in this paper to "characterize" 
tornadoes based on these two parameters. Ideally, each storm should also be charac- 
terized by three-dimensional meteorological parameters such as temperature, humidity, 
pressure, wind, as well as funnel size, shape, duration, etc. It will be years before we 
are able to measure these meteorological parameters and their time changes accurately. 
The term "characterization" is used in this paper to express specific storm characteris- 
tics. 

As a test characterization of tornadoes, Palm Sunday storms were selected 
because the author took a large number of aerial photographs to complete a report by 
Fujita, Bradbury, and Van Thullenar (1970). Since tornado characterization was not of 
primary significance when the initial studies were completed, all aerial i*iotographs and 
notes were reexamined to establish both intensity and area classifications of the 25 major 



18 



storms for this day that were surveyed by the author. For unsurveyed storms, Storm 
Data were examined with the help of topographical maps to determine the most reasonable 
F md leg a scales. After extending the characterizations to include April 10, the 
pre-Palm Sunday storm day, the frequencies of occurrence belonging to each intensity and 
area were plotted as a function of the initial time of tornado occurrences (see Fig. 7). 
This figure clearly shows that the center of gravity of frequencies moved from F 1 on 
April 10 to F 2 on the 11th, suggesting that Mother Nature does produce intense torna- 
does without increasing simultaneously the number of small ones. This means that the 
formation of an intense tornado is not the result of an accidental growth out of an 
increased number of small ones. Apparently, differing meteorological conditions produce 
tornadoes of differing intensity and area. Similarly the center of gravity in the Meso 
category on the J 0th shifted to that of Macro on Palm Sunday. Note that Macro-sized 
tornadoes formed literally one after another between 12 noon and midnight. 

In order to learn more about the distribution of tornado area and intensity, Tecson 
(1971), made a test analysis of 893 tornadoes in 1965 based on the storm description given 
in Storm Data. His results revealed that 77% of these tornadoes are in FO and F 1 
categories while F3 and F4 storms are only 5% and 1%, respectively, of the total num- 
bers , revealing that the frequency of intense tornadoes is very small. 

When a similar analysis of Japanese tornadoes was made based on Japanese Storm 
Data ( 1950-69 ) 80% turned out to be in FO and F 1 categories while 3% were F3. There 
were no reports of F 4 or higher categories. These results when examined along with 
U. 3. data, as shown in Fig. 8, revealed that the distribution of FO, Fl, and F2 of 
Japa^ise and U. S. tornadoes are very similar to each other. The real difference is the 
fact that F 4 and more intense storms occur in the U. S. but not in Japan. 

Significant differeiues are also found in the frequency distribution of macro, giant, 
and decagiant categories which are practically non-existent in Japan. These storms are 
most likely to develop under very specific meteorological conditions experienced exclu- 
sively in the midwestern United States. In a recent five year period, the number of 
tornadoes reported in Japanese Storm Data was 13 (273) in 1965, 15 (315) 1966, 7 (147) 
1967, 13 (273) 1968, and 15 (315) in 1969. Numbers in parenthesis indicate frequencies 
prorated to the U. S. area which is 21 times larger. These prorated frequencies are 
comparable to U. S. frequencies in 1951-52 when tornado frequencies began to increase 



19 



APR 10 (PRE -PALM SUNDAY), 1965 



-6om to 6 am TORNADO DAY 



APR II (PALM SUNDAY), 1965 




Fig. 7. Intensity and area distribution of pre- and Palm Sunday tornadoes of 1965. The 
numbers indicate the frequencies of characterized tornadoes within 2 -hour period of each 
tornado <liy defined as a 24-hour period from 6 a. m. to 6 p. m. This definition of tornado 
day is reasonable especially when the tail end of tornado activity extends to early in the 
morning. 



FREQUENCY OF TORNADOES BY INDIVIDUAL AREA AND INTENSITY 



JAPAN 
20 yeors (1950-69) 

156 Tornadoes 



AREA 



42% 



30 



T« 6m mi 



21 



ME 5S" Gt 05" 



INTENSITY 

40 40 



17% 



FO Fl F2 F3 F4 F5 F6 



UNITED STATES 

One year (1965) 

893 Tofnodoes 



AREA 




PALM SUNDAY 

On^ ., (II Apr 1965) 
47 Tornadoes 



AREA 



53% 



15 15 



TR OM Ml ME MA Gl 



m^ 



INTENSITY 



zrv. 



26 26 



17 



FO Fl fl F3 F4 F5 F6" 



Fig. 8. Typical distribution of tornado frequencies as functions of both area and intensity. 



20 



due mainly to the improvement in reporting systems. It is very likely that Japanese-type 
distribution is af^licable to most odier island countries such as New Zealand, England, 
Italy, etc. , ¥^ere local weather bureaus have not been predicting tornadoes successfully. 
However, meteorological conditions giving rise to the formation of weak and/or small 
U. S. tornadoes could be found in other parts of the world if we look for them. For the 
improvement of U. S. local forecasts, investigation of unique ccHiditions associated with 
large and strong tornadoes is of vital importance. 

Further apidicati(» of intensity and area cfaaracterizati(xi appears in Fig. 9. The 
figures on the left side show that autimin is the tornado season in Japan. It is seen that the 
intensity of tornadoes decreases in the spring as well as their monthly occurrences. 



ANNUAL AND DIURNAL VARIATION OF JAPANESE TDRNADOES. 1950-69 




JM FEB MM AmiWr JUNJULMJGSEPOCTNOrOEC 



Fig. *?. Variation of characterized frequencies of Japanese tornadoes in 20-year period, 
1950- 1%9. Unlike U.S. tornadoes, autumn is the tornado season with a significant minimum in 
April. Intense. F 2 and F 3 tornadoes are most frequent in September and October when large-area 
storms are also expected. Note that diurnal variation of Japanese tornadoes is insignificant. 



21 



Diurnal variation (right side) shows an overall maximum around local noon, but the strong 
ones, F2 and F3, remain active nearly 24 hours a day. It is of interest to find tiiat the 
maximum frequoicy of Meso tornadoes, the largest category with die exception of one 
Macro tornado, occurred b^ween 8 and 9 in the morning. The intensity also shows a 
maximum at this time. Meteorological reasons for these occurrences have not been 
explored. 

8. APPLICATICHM OF F SCALE TO THE SURVEY OF HURRICANE DAMAGE 

Unlike tomadic storms, hurricanes and typhoons are long lived, covering and 
affecting large areas. From meteorological points of view these storms have been charac- 
terized by their parameters vrtiich are measured by satellite, radar, vertical soundings, 
aircraft, and other means of observations. Moreover, most meteorological instruments 
are designed and constructed to withstand most hurricanes and typhoons. 

Since measured winds are more accurate than estimated F-scale winds, it is not 
necessary to determine F-scale winds if nearby anemometers are available. In reality, 
however, the number of anemometers existing inside a vast area of storm damage is so 
inadequate that both ground and aerial surveys are required to determine meso to micro- 
scale damage patterns caused fay typhoons and hurricanes. F-scale estimates inside 
hurricane areas are, therefore, very useful in establishing the patterns of damaging wind, 
^ich cannot be determined t^ using only a limited number of anemcHneters. 

Presented in Fig. 10 is the distribution of F-scale winds inside hurricane Camille 
of August 17-18, 1%9 when the storm tra\^lled inland. F-scale winds and their direc- 
tions were determined by the author from an aerial survey a few days after the storm. 
There were only several wind recorders within the damage area, making It very difficult 
to establish damage patterns based exclusively on measured wind speeds. 

It should be noted that the RECON winds inside Camille prior to the landing were 
about F4 equivalent, which was considerably stronger than F3, the highest value of die 
F scale wind estimated near the storm's landing site. Such a difference could be related 
to the weakening of the storm, the nature of REOON and anemometer winds, etc. 

It is worthA^lle, at this point, to reexamine various definitions of wind speeds so 
as to inter-relate them as much as possible. Strictly speaking, however, there is no way 
of converting one tjrpe of wind into others because the lime variation of wind speed cannot 

22 




Fig. 10. Distributitm of F-scale damage caused by hurricane Camille of 
August 17-18. 1969. 

be expressed as a simple periodic function. The following terms are used in this paper 
to express the nature of the measured wind. 

# FASTEST 10-MIN WIND or Maximum 10-min Wind 

The maximum wind speed averaged over any ten -minute period 
at a given station during a specified period. 



# FASTEST l-MDSf WIND or Maximum 1-min Wind 

The maximum wind speed averaged over any one-minute period 
at a given station during a specified period. 

23 



# PEAK GUST or Maximum Instantaneous Wind 

The highest instantaneous wind speed at a given station within 
a specified period. 

# FASTEST MILE WIND 

The maximum speed of any mile -passage of wind at a given 
station during a specific period. The averaging period decreases 
inversely proportional to the fastest mile wind speed. 

# FASTEST 1/4 MILE WIND or F-scale Damaging Wind 

Ihe maximum speed of any 1/4 mile-passage of wind at a given 
station during a specified period. The averaging period decreases 
inversely proportiooal to F-scale wind speed. 

Because these five wind measurements are not always available simultaneously 
from each station affected by a specific storm, it is desirable to establish certain rela- 
tionships between these winds, with the understanding that the relationships are approxi- 
mate. 

It is customary to express the peak gust as a function of both mean wind speed and 
gustiness factor, thus 

VpG = (1+79 ) Vioorl (5) 

where V pq denotes the peak gust, q the gustiness factor defined by 



gust speed - lull speed 
mean wind speed * 

and Vio and V| , die 10 -min and 1-min mean speed, respectively. Ihe gustiness 
factors vary widely from storm to storm as well as the location of anemcHneters. As 
shown in a scatter diagram of Fig. 11, the gustiness factors of most tyi^ooos in Japan 
are between 0. 3 and 1.5, indicating that the range of lull to gust speeds is comparable 
to the mean speed. 



24 




Fig. 11. Scatter diagram indicating a linear 
relationship between the peak gust and the 
fastest mean wind of Japanese typhoons. Ihe 
gustiness factor, q , can be approximated as 
1.0. Data points indicated by black dots rep- 
resent the values measured within Ise-wan 
lyphoon of 26 September 1959. The number of 
dead and missing due to this hurricane was 
about 5000. 



\(o. FASTEST lO-MIN WINO- 



As a first approximation we express the wind speed by a sinusoidal gust super 
imposed upon the mean wind velocity V , thus 

27rt 



V = f(t) = V + -^ g V cos p 



(6) 



where p denotes the period of the sinusoidal gust (see Fig. 12). So long as the period p 
is much shorter than the averaging period, one or ten minutes, the averaging period does 
not alter the mean value. Eq. (5) can thus be written simply ' j 



P6 



= (1 + T9) V . 



(7) 



The fastest 1/4-mile or F-scale damaging wind can be obtained by integrating the 
wind passage centered at the time of the peak gust. Namely, we write 



25 



Q • 


6USTINESS FACTOR , 










-yAf +jAf 










•—At — • 








r 


xzTix:";: , 


/"^ 




^ ^V. 


^m. 


1 
V 


■ ■ ■■ 


\ / »-* \T 




\ 








\ / "^ \ / 




\ 




\ 1 




V/.-.L-..X/ 


/FASTEST. 


z 

i 

UJ 

-1 




i 

O 
Ul 
UJ 


v^ 


o 




;. .f>ATH:-: 


Z 


1- 


Q. 
(O 




^ 






CO 


(9 


o 
z 




z 






UJ 


af 






4 






1- 


4 


^ 




UJ 

2 






(O 

If 


UJ 

a. 


<^ 




•k 






• 








l> 






W 


o 

a. 


M 










> 


> 


> 








::•■•• ■■:•■" 











Fig. 12. Time variation of a simplified gusty wind. The wind speed as a function of time is expressed by a 
sinusoidal variation superimposed upon a constant wind. 



Vc At 



= \ mile = j 



gpv 



Vdt 
Ztt 



+ TAt 



«- P J-TAt 



= VAt + -T — I sin „ 

wA. ■ 9P^ irAt 
= V At + -:; sin 

2 IT D 



(8) 



where At is the time in which the F-scale wind travels throu^ a distance of 1/4 mile 
at the rate of Vp , the speed of the fastest 1/4 mile wind. 
Substituting At in Eq. (8) by -^ , we have 



or 



- mile = -^ 



V = ( — + 
V Vf 



qpV 
Zir 

2gp 



sin 



4pVf 



sin 



4pVF 



) 



-I 



(9) 



From this equation, we are able to convert F-scale wind speed, Vp , into the mean wiml 
speed, V. It should be noted, however, that both gustiness factor, g , and period, p , 
must be known for the conversion. 

In special cases when Vp is extremely high or low, Eq. (9) can be reduced 



26 



simply to 

Vf = V for very low wind speeds (10) 

and 

Vf =V(l + -2g) for very high wind speeds ( 11 ) 

because the sinusoidal term in Eq. (9) becranes small in compariso': with the other term 
and may be neglected when Vf is small and the sinusoidal term can be approximated as 
"2 g / V F for large values of Vp . 

The above approximation implies that die damage caused by an extremely hi^ wind 
is closely related to the peak-gust speed as expressed by Eqs. (7) and (11). On the other 
hand, weak damage due to low winds, such as 40 m^i, is a result of the time integrated 
stress of repeated weak gusts and a steady flow of air against weak structures. 

Two gustiness parameters, g and p, are closely related to the turbulent charac- 
teristics of damaging winds which are usually higjily gusty near the ground. For hurri- 
canes and typhoons the gust period varies between the order of seconds and minutes. For 
tornadoes, however, little is known regarding their gustiness characteristics. A tornado 
wind trace recorded by the Tecumseh Health Study, University of Michigan and reported 
by Fujita et al. (1970) in their Paim Sunday Tornado paper, showed a 151 mph peak gust 
characterized by a gust period of about 20 seconds . 

Coming back to the problem of converting F-scale wind speeds into anemometer- 
measured wind speeds, it should be emphasized that the mean wind speed represents 
F-scale wind speed when the speed is very low and that F-scale speed approaches the 
peak gust speed as the speed increases. Figure 13 shows two dotted straight lines repre- 
senting the mean and peak gust speeds. In computing the mean speed as a function of 
F-scale speed using Eq. (9), two gustiness periods, 15 sec and 30 sec, were assumed. 
These curves in Fig. 13 shown in short and long dashed lines respectively, reveal, as 
expected, that the F-scale wind is very close to the peak gust when the speed is in excess 
of 150 mph. This is because the maximum damage occurs mostly at the time of the peak 
gust. As the wind speed decreases, both dashed lines approach the mean spieed line in 
damped oscillatory manners because Eq. (9) includes a sinusoidal term. By specifying a 
number of gustiness periods, the same number of dashed lines are to be added in the 
figure. 



27 



40 



73 



113 



158 



207 



261 



F SCALE 



-200 



N 

Z 

u 

X 



-ISOmph 



< 



l> 



-lOOmph 



319 mph 




7-Vf conversion function. Eq. (12) 



i + ig 



ISO 200 ^0 

Vf, fastest 1/4 MILE WIND 



300 mph 



Fig. 13. A function to convert fastest 1/4 mile wind into mean wind and pealc gust. 



Since it is impractical to produce conversion functions corresponding to a number 
of gustiness periods, a specific gustiness period was selected for the conversion purposes. 
The period was selected so that the conversion curve connects smoothly the 40 mjrfi or 
F wind speed with the peak gust speed. A heavy line departing from V = Vp = 40 mph 
represents the proposed conversion function (see Fig. 13). The gustiness period for this 
conversion function can be obtained by selecting the angles to reduce the sinusoidal term 
in Eq. (9) to zero. These angles are 0, ir , 2v , . . . mr . In view of the fact that 
represents Vp = infinity, the next larger angle, v should be used for this purpose. 
Thus we equate ir with the angle in Eq. (9), thus 



ir - 



4pVF 



or p = 



4Vf 



28 



in mph units. Putting Vf = 40 mj* as originally assumed, we have 

o 1 , 3600 -. _ 

^ = mo ^°"^ = "T60" ^^"^ = 22.5 sec. 

The equations for converting Vp into V and Vpq are thus expressed in mph by 

V = —. Az (12) 



1 . 45g . 
Vf ^ . ''' 

1 + ig 


tr 


90 Vf 


^ + ^^^ .in 


IT 


Vp ^ TT ^'" 


90 Vp 



and VpG = — i T= (13) 

1 45g . w 

Yp ^ T ^'" -9ov7 

These equations can readily be used in relating F-scale wind speed, Vp , with both 
fastest mean speecfe and peak gust speeds measured and/or recorded by anemometers. It 
should be emphasized that the period of the fastest mean speed, averaged over one- or 
ten -minutes, must be selected as being close to the time of the occurrence of the fastest 
1/4 mile wind. Such a selection will permit us to approximate actual winds as a simple 
sinusoidal function expressed by Eq. (6) and Fig. 12. 

9. USE OF TABLE CONVERTING MEASURED WINDS INTO F SCALE WINDS 
FOR HURRICANES 

Unlike wind records within a tornado, a large number of measured maximum winds 
is available within a specific hurricane during its life time. These winds are reported in 
forms of FASTEST 10- or 1-MINWIND, PEAK GUST, FASTEST MILE WIND, RECON 
WIND, and others. 

In orcter to relate these winds with detailed patterns of F-scale winds, which can 
be mapped with isolines of FO, Fl, F2, etc. , it is necessary to convert these measured 
winds into F-scale wind. Althou^ the F scale estL lates from structural and/or tree 
damage are just enou^ to distinguish the stepped F-scale values, measured wind speeds 
are to be converted into the fractional scales such as FO. 3, F 2. 3, F 3. etc. , thus 
allowing the determination of more accurate values at anemometer locations but not 
everywhere over the areas of wind damage caused by a storm. 

Table IV, computed from Eqs. (12) and (13), permits us to determine fractional 
F-scale values from measured wind speeds at 2 m/s intervals given in three units, m/s. 



29 



Table IV. TABLE TO ESTIMATE FUJITA SCALE FROM RECON WIND, PEAK GUST, AND MEAN WIND. 
Gustiness factor, g, may be assumed 1.0 for most land stations, g = 1.5 can be applied 
to very gusty winds often observed at inland stations and g = 0.5 to light gust con- 
ditions. To determine F scale, first estimate gustiness factor at the top, then look 
down Che appropriate column to the measured wind speed given in m/s, kt, and mph. 



MEASURED WIND SPEED 


RECON 


WIND 






PEAK GUST 










MEAN WIND 




m/s 


kt 


mph 


g= 


=0.0 


g= 


=0.1 


g= 


=0.5 


g=1.0 


g= 


=1.5 


g= 


=0.5 


g=1.0 


g=1.5 


18 


35 


40 


F 


0.0 


F 


0.0 




- 




- 




- 


F 


0.0 


F 


0.0 


F 0.0 


20 


39 


45 


F 


0.1 


F 


0.1 




- 




— 




— 


F 


0.2 


F 


0.3 


F 0.4 


22 


43 


49 


F 


0.3 


F 


0.3 




- 




- 




- 


F 


0.4 


F 


0.6 


F 0.8 


24 


47 


54 


F 


0.4 


F 


0.4 


F 


0.0 




- 




- 


F 


0.6 


F 


0.8 


F 1.1 


26 


51 


58 


F 


0.6 


F 


0.6 


F 


0.2 




- 




- 


F 


0.8 


F 


1.0 


F 1.4 


28 


54 


63 


F 


0.7 


F 


0.7 


F 


0.4 


F 


0.0 




- 


F 


0.9 


F 


1.2 


F 1.6 


30 


58 


67 


F 


0.8 


F 


0.8 


F 


0.6 


F 


0.3 




_ 


F 


1.1 


F 


1.4 


F 1.8 


32 


62 


72 


F 


1.0 


F 


1.0 


F 


0.7 


F 


0.5 


F 


0.0 


F 


1.2 


F 


1.6 


r- 2.0 


34 


66 


76 


F 


1.1 


F 


1.1 


F 


0.8 


F 


0.6 


F 


0.3 


F 


1.4 


F 


1.8 


F 2.2 


36 


70 


81 


F 


1.2 


F 


1.2 


F 


1.0 


F 


0.8 


F 


0.6 


F 


1.6 


F 


2.0 


F 2.4 


38 


74 


85 


F 


1.3 


F 


1.3 


F 


1.1 


F 


0.9 


F 


0.8 


F 


1.7 


F 


2.1 


F 2.6 


40 


78 


90 


F 


1.4 


F 


1.5 


F 


1.2 


F 


1.1 


F 


1.0 


F 


1.9 


F 


2.3 


F 2.8 


42 


82 


94 


F 


1.5 


F 


1.6 


F 


1.3 


F 


1.2 


F 


1.1 


F 


2.0 


F 


2.5 


F 3.0 


44 


86 


98 


F 


1.7 


F 


1.7 


F 


1.5 


F 


1.4 


F 


1,3 


F 


2.1 


F 


2.6 


F 3.1 


46 


89 


103 


F 


1.8 


F 


1.8 


F 


1.6 


F 


1.5 


F 


1.4 


F 


2.2 


F 


2.8 


F 3.3 


48 


93 


107 


F 


1.9 


F 


1.9 


F 


1.7 


F 


1.6 


F 


1.5 


F 


2.4 


F 


2.9 


F 3.4 


50 


97 


112 


F 


2.0 


F 


2.0 


F 


1.8 


F 


1.7 


F 


1.7 


F 


2.5 


F 


3.1 


F 3.6 


52 


101 


116 


F 


2.1 


F 


2.1 


F 


1.9 


F 


1.8 


F 


1.8 


F 


2.6 


F 


3.2 


F 3.8 


54 


105 


121 


F 


2.2 


F 


2.2 


F 


2.0 


F 


1.9 


F 


1.9 


F 


2.7 


F 


3.3 


F 3.9 


56 


109 


125 


F 


2.3 


F 


2.4 


F 


2.1 


F 


2.1 


F 


2.0 


F 


2.9 


F 


3.5 


F 4.1 


58 


113 


130 


F 


2.4 


F 


2.5 


F 


2.3 


F 


2.2 


F 


2.1 


F 


3.0 


F 


3.6 


F 4.2 


60 


117 


134 


F 


2.5 


F 


2.6 


F 


2.4 


F 


2.3 


F 


2.2 


F 


3.1 


F 


3.8 




62 


120 


139 


F 


2.6 


F 


2.7 


F 


2.5 


F 


2.4 


F 


2.3 


F 


3.2 


F 


3.9 




64 


124 


143 


F 


2.7 


F 


2.8 


F 


2.6 


F 


2.5 


F 


2.4 


F 


3.3 


F 


4.0 




66 


128 


148 


F 


2.8 


F 


2.9 


F 


2.7 


F 


2.6 


F 


2.5 


F 


3.4 


F 


4.1 




68 


132 


152 


F 


2.9 


F 


3.0 


F 


2.8 


F 


2.7 


F 


2.7 


F 


3.6 


F 


4.2 




70 


136 


157 


F 


3.0 


F 


3.1 


F 


2.9 


F 


2.8 


F 


2.8 


F 


3.7 








72 


140 


161 


F 


3.0 


F 


3.2 


F 


3.0 


F 


2.9 


F 


2.9 


F 


3.8 








74 


144 


166 


F 


3.1 


F 


3.3 


F 


3.1 


F 


3.0 


F 


3.0 


F 


3.9 








76 


148 


170 


F 


3.2 


F 


3.4 


F 


3.2 


F 


3.1 


F 


3.1 


F 


4.0 








78 


152 


175 


F 


3.3 


F 


3.5 


F 


3.2 


F 


3.2 


F 


3.2 


F 


4.1 








80 


155 


179 


F 


3.4 


F 


3.6 


F 


3.3 


F 


3.3 


F 


3.3 


F 


4.2 








82 


159 


183 


F 


3.5 


F 


3.7 


F 


3.4 


F 


3.4 


F 


3.4 












84 


163 


188 


F 


3.6 


F 


3.8 


F 


3.5 


F 


3.5 


F 


3.4 












86 


167 


192 


F 


3.7 


F 


3.8 


F 


3.G 


F 


3.6 


F 


3.5 












88 


171 


197 


F 


3.8 


F 


3.9 


F 


3.7 


F 


3.7 


F 


3.6 












90 


175 


201 


F 


3.9 


F 


4.0 


F 


3.8 


F 


3.7 


F 


3.7 












92 


179 


206 


F 


4.0 


F 


4.1 


F 


3.9 


F 


3.8 


F 


3.3 












94 


183 


210 


F 


4.0 


F 


4.2 


F 


4.0 


F 


3.9 


F 


3.9 












96 


187 


215 


F 


4.1 


F 


4.3 


F 


4.1 


F 


4.0 


F 


4.0 












98 


190 


219 


F 


4.2 


F 


4.4 


F 


4.1 


F 


4.1 


F 


4.1 













30 



kt, and mps. Three types of winds, RECON, PEAK GUST, and MEAN WINDS, are tabu- 
lated because they are usually reported to express hurricane winds. To determine the 
F scale, first estimate the gustiness factor given at the top, then follow the appropriate 
column down to the measured wind speed given in the left three columns. 

RECON winds are one of the most important parameters to estimate the intensity 
of hurricanes over the ocean. An example of Doppler winds measured at 1770 ft through 
the eye of hurricane Gladys of 17 October 1968 is shown in Fig. 14. Most data points, 



DOPPLER WINDS AT 1770 FEET , HURRICANE GLADYS 

South -* North Penetrotion , 17 October 1968 




Fig. 14. Doppler winds measure-l at 1770 ft through the eye of hurricane Gladys on October 17, 1968. 
Winds were measured by RFF's DC6B while traversing south to north during the research mission. 



plotted for every 5 -sec, are within 0. 95 and 1. 05 of the center values about which the 
wind fluctuates. F-scale winds in g = 0. 1 column were computed from Eq. (12) assuming 
that RECON wind represents the center value. If the gustiness factor is zero, the left 



31 



column for g = 0. should be i-sed for conversion of such RECON winds inio F-scale 
values. It is likely that gustiness factor for RECON winds is much smaller than that of 
turbulent wind recorded at the levels of structures on the ground. 

Both PEAK GUST and MEAN winds are tabulated for three gustiness factors, 0. 5, 
1.0, and 1. 5. For most wind stations g = 1. may be used for practical purposes. For 
very gusty wind, g = 1. 5 will result in a better approximation. With this gustiness factor, 
the lull to gust speed will vary from 0. 25 to 1. 75 times the mean wind speed. For insig- 
nificant gust cases which are likely to occur at coastal stations or land stations at night, 
the use of g = 0. 5 is often appropriate. 

Some stations report both PEAK GUST and the corresponding MEAN WIND. Each 
of these values can be used to determine the F scale independently by assuming a gustiness 
factor. Two F scales thus estimated independently could differ significantly because the 
gustiness factor assumed may not be accurate enou^. If they differ beyond an allowable 
limit of 0. 2 to 0. 3 in F scale, we may use the mean values of the two estimated F scales. 
For example, if reported hurricane winds from a station are 130 mph PEAK GUST and 
90 mpti MEAN WIND, Table IV gi F scales corresponding to these values as being 

F 2.3 and t i.9 for g = 0. 5 
F 2. 2 and F 2.3 for g = 1.0 
F 2.1 and F 2.8 for g= 1.5 , 
respectively, indicating that g = 1.0 turns out to be the best approximation because F2. 2 
and F 2. 3, estimated respectively from the peak gust and the mean wind, show very good 
agreement. If we compute the average F scales obtained by assuming g = 0. 5 and g = 1. 5 
they are F2. 1 and F2. 4, respectively and each of these averages are close enough to 
F 2. 2 and F 2. 3 which would represent the appropriate F scale at this station. Neverthe- 
less, the most reasonable estimate of F scale from Table IV can be performed by 
selecting a gustiness factor, g , which would minimize the difference in the F scales 
estimated independently from both peak gust and maximum mean speed occurring at a 
station during a specific hurricane. 

Another example of F scale estimate from an anemometer record is shown based 
on a gust -recorder trace from Reynolds Metal Company located just north of the track of 
hurricane Celia of 1970. As shown in Fig. 15 the peak gust and the maximum mean wind 
were 138 mph and 110 mi* ahead of the eye and 134 mi* and 100 mph to the rear of the 



32 




oot MifaaV 4.W ST" 



zu MMusT s. i9ro za 



zzz 



zu 



Fig. 15. Wind speed trace <rf Hurricane Celia. August 3-4. 1970 recorded at Reynolds Meul Company. 



eye. These values result in the best possible F scale Y/hea q = 0. 5 is applied. F scales 
for this storm before and after the passage of me eye are thus estimated to be F2. 5 and 
F2.3, respectively. 

Rejmolds Metal Company was located to the rig^t of Celia's center, where the 
St -'.gest hurricane winds are expected. Simpscm's (1970) aerial survey and subsequent 
investigation revealed that the storm damage occurring predominantly on the left side of 
the track was caused by strong westerly winds. The westerly winds were estimated to be 
stronger than those on die other side, causing ccmsiderable damage in Corpus Christi. 
Since dynamical aspects of the hurricane circulation resulting in such damage are quite 
unusuil, a joint research between Dr. Simpson and the author was initiated and fine 
patterns of damage streaks are being obtained. 

In order to demonstrate structural damage in relation to the pattern of F-scale 
mnd, distributi(Hi of damaged houses reported by Ishizaki et al. (1961) was combined with 
F-scale winds. As shown in Fig. 16, the Ise-wan ty{Aoon packed with 150 mph winds in 
Che eastern sector of the 930 mb center landed near the southern tip of Kii peninsula 
shortly before 1900 Japan Standard Time. During the next 3 hours a tongue of up to F 2. 9 
winds moved north, resulting in a 3. 6-m or 12-ft scorm surge and 5,000 fatalities. In 



33 




Fig. 16. Relation between the F-scale wind and the relative number <rf houses demolished 
by the Ise-wan Typhn«i of 26 September 195^ which diagonally crossed the central part of 
Japan. 



Nagoya City 6,569 houses out of 252, 145 were demolished due mostly to strong wind. 



10. CHARACTERIZATION OF HURRICANES AND TYPHOONS 

For the purpose of determining the statistical differences in the maximum wind 
speeds of Pacific tyi^oons, Atlantic hurricanes, and Pacific hurricanes, maximum wind 
speeds of all storms tabulated in the National Summary, Climatological Data (1960-69) 
were chosen to obtain cumulative numbers of storms as a function of F scale converted 
from the published maximum speeds which are based on various ts^pes of measurements. 
The result summarized in Fig. 17 shows that 34 Pacific hurricanes leveled off before 



250 — 



2001— 



(T 
O 

I- 
(/) 

u. 150 
O 

(T 
tiJ 

ca 

5 100 



50 



OL 



CUMULATIVE NUMBER OF TROPICAL CYCLONES 

lO-year period, 1960-1969 



90%(F33) 



507o(F20) 




211 RftClFiC TYPHOONS 



61 ATLANTIC HURRICANES 



Comille 



34 WVCIFIC HURRICANES 



50%(FI3) , 

I L 



90%(F2 3) 



10 



15 



20 



25 



30 



35 



40 



Fujita Scale 



Fig. 17. Cumulative number of tropical cycl<Mies during a 10-year period. 1960-1%9, plotted 
against Fujiu scale wind speed. 



reaching F3.0, while Atlantic hurricanes, 61 in total number, kept increasing their 
cumulative number until Camille of 1969 hit F4.0. 

Pacific typhoons occurred almost 3. 5 times more often than hurricanes but the 
maximum F scale was very close to that of Camille. The F scale corresponding to the 



as 



90% cumulative numbers vary among these species of storms. Namely, 90% of Pacific 
hurricanes are less than F2. 3, Atlantic hurricanes, less than F2.8, but 90% of Pacific 
tyjrfioons may be up to F 3. 3. 

In an attempt to d^ermine the relationship between the central pressure and 
F-scale wind speed, a scatter diagram (see Fig. 18) vrais made by plotting the central 







970- 

9€0- 

950- 

940- 

930- 

920- 

9»0 

900- 

890- 

880- 

870 



Sh^ 



100 

_J 



150 
I 



200 m|^ 
I 



iO 






\ \ "".^ 






10 -YEAR PERIOD, 1960-1969 







'^^\ \, 



'^ 



• PACIFIC TYPHOONS 

• ATUVNTIC HURRICANES 
- PAQFIC HURRICANES 



liiiiliiiii I ■ I I I I I I I 1 1 I I I I — I I I 1 — LvJ 1 1 J^iJ 1 1 1 L_M 



FOO 



FIO 



F20 



F30 



P4C 



Fig. 18. A scatter diagram showing the central pressures of hurricanes and typhoons plotted 
against F-scale wind speed. Note thit three species of storms, each enclosed by an elongated 
ellipse, are slightly different. For a given wind speed, central pressures of Racific typhoons in 
average are lower than hurricanes. 



pressures of 1960-69 storms against 

V2 = 39.7(F + 2)^m*/sec*=199(F + 2)'mph' 



(14) 



which is obtained as the square of Eq. (2). 

If hurricanes and typhoons are assumed to be Rankine-type vortices with cyclo- 
stroi*iically balanced wind everywhere, the pressure deficit at the center is given by 



36 



AP = 2. lpV.2 (15) 

where p is the density of air assumed to be constant and V« , the speed of die maxi- 
mum wind around the eye wall. 

Figure 18 reveals, however, that die "pressure-deficit coefficient", 17 defined 
by 

A P = V \ P^^» 

where p = 1. 16 kg/m^ at 1010 mb and 30° C virtual temperature or about 27*'C air 
temperature, varies widely between 1 and 5 or more. A slight variation of p , which 
was neglected in obtaining Eq. (15), does not produce such a variation. Moreover, the 
average scatter for each storm group shows a successive shift from Pacific typhoons to 
Atlantic hurricanes to Pacific hurricanes. This means that F- scale maximum wind speed 
as well as the radial distribution of tangential wind speed are needed for an improved 
characterization of tropical cyclones. 

If we approximate tangential wind speeds inside and outside the circle of die maxi- 
mum wind, respectively, by 

V = ko r where ko = V« / r» (inside) 
and V = k^r"" where k^ = rjf V, (outside) , (16) 

cyclostrophic approximation will permit us to write 

^ = -^o + -^b (17) 

where "^o = — and ''7b = -r 
a b 

and a is estimated to be larger than 1 and b is smaller than 1 and close to 0. 5 
according to Ridil (1954). By selecting proper values of a and b , we will be able to 
characterize bodi Pacific and Atlantic storms in terms of intensity and radial distribution 
of wind. 

TTiese evidences imply that the central pressure of hurricanes and typhoons can- 
not be related to the maximum F- scale wind speed without accepting a large standard 
deviaticHi. For a given F-scale wind, the central pressure seems to vary as much as 
50 mb throughout the range of hurricane F-scale winds. If such variations are causec* by 
the deviation of hurricane aiKi t5rirfioon vortices from simplified Rankine vortices , proper 
interpretation of "J? , taking into consideration the dynamical aspects of the circulation 
as well as Coriolis force, will be of vital importance. 



37 



Nonetheless, the scatter diagram of Fig. 18 clearly shows major statistical 
differences between Atlantic hurricanes, Pacific hurricanes, and Pacific typhoons which 
are to be characterized by various parameters. 

11. CONCLUSIONS 

Numerous severe storms such as hailstorms, tornadoes, tropical cyclones, etc. 
are spawning and dying out after minutes or days of their life given by Mother Nature. 
Unlike human beings or others in the animal world, individual storms belonging to one 
species are so different in size and intensity that each must be characterized properly in 
order to assess their behavior and effects on human life. 

Investigations of U. S. and Japanese Storm Data revealed that the area affected 
by an individual storm reported as a tornado or tatsumaki is less than 0. 001 sq. mi. , 
v^ile the largest one in U. S. was in excess of 100 sq. mi. , or 1:1,000,000 in areal ratio. 
The range of the maximum wind spped inside storms reported in Storm Data as tornadoes 
varies between less than 73 mph, Beaufort 12 and up to about 300 mpti. It is misleading 
to assess tornado activities relying heavily on their occurrences without describing 
individual characteristics. 

In order to avoid the existing possibility that a tornado affecting 0. 001 sq. mi. and 
the other affecting a 10 sq. mi. arta are treated with equal weight in statistical analyses, 
the author proposed to categorize individual tornado area according to its logarithm. The 
maximum wind speeds inside tornadoes are also categorized by F-scale which was devised 
to connect the upper end of Beaufort force with the low end of Mach number. 

Several test analyses of tornado area and intensity thus defined now appear to be 
very useful. A comparison of Japanese and U. S. tornadoes revealed that 75% of all 
tornadoes are similar in area and intensity distributions . The only difference is that U. S. 
hao extremely large and/or strong tornadoes which do not exist in Japan. A further 
investigation of tornadoes in other parts of the world will probably result in similar dis- 
tributions . 

An initial attempt to investigate hurricane and typhoon damage with F-scale 
categories was made. The fact that the maximum F-scale damage expected in hurricanes 
and typhocais reaches F 3 permits the identification of damage into a maximum of three 
categories. This turned out to be very useful in estimating the fine structure of storm 

38 



circulation which cannot be studied otherwise. 

Although this has been the first attempt to establish and identify storm character- 
istics by numbers obtained through an "educated guess", preliminary application presented 
in this paper revealed the potential v^alue of the concept of "characterization". 

ACKNOW LEDGEMENT 

The author is very grateful to Messrs. Vincent J. Oliver and Linwood F. Whitney, 
Jr. , of the National Environmental Satellite Service, Drs. Robert H. Simpson of the 
National Hurricane Center, R. Cecil Gentry of the National Hurricane Research Labora- 
tory, Edwin Kessler of the National Severe Storms Laboratory, and Mr. L. A. Joos, 
Regional Climatologist, Central Region, National Weather Service Office, for their com- 
ments during the drafting stage of this paper. 

The author is also grateful to the staff members c* the Satellite and Mesometeorol- 
ogy Research Project of the University of Chicago, espev dUy to Jaime J. Tecson and 
Dorc^y L. Bradbury for their assistance toward the completion of this paper. 



39 



REFERENCES 

Battan, L. J. (1959): Duration of Tornadoes. Bull. Amer. Met. Soc. , 40, 340-342. 
Beebe, R. C. (1960): The Life Cycle of the Dallas Tornado. Research Paper No. 41, 

U.S. Weather Bureau, 3-52. 
Brown, R. A. and T. T. Fujlta (1961): Report on the Oiicago Tornado of March 4, 1%1, 

13 pp. 
Clarke, R. H. (1%2): Severe Local Wind Storms in Australia. Div. of Meteor. Physics 

Tech. Paper No. 13, Commonwealth Scientific and Industrial Research Or^niza- 

tion, Melbourne, Australia, 56 pp. 
Climatological Data 1960-69: National Summary. U.S. Weather Bureau, NWRC, 

Asheville, N. C. 
Court, A. (1970): Tornado Incidence Maps. ESSA Tech. Memorandum ERLTM-NSSL 49, 

NSSL, Norman, Okla. , 76 pp. 
Darkow, G. L. and J. C. Roos (1970): Multiple Tornado Producing Thunderstorms and 

their Apparent Cyclic Variations in Intensity. Preprints of 14th Radar Meteorology 

Ck)nference, Tucson, Arizona, 305-308 
Flora, S. D. (1953): Tornadoes of the United States. University of Oklahoma Press, 

Norman, Oklahoma, 221 pp. 
Fujita, T. T. (1960): A Detail Analysis of the Fargo Tornadoes of June 20, 1957. 

Research Paper No. 42, U, S. Weather Bureau, 67 pp. 
Fujita, T. T. (1963): Analytical Mescxneteorology: A Review. Meteorological Monograpdi 

5, No. 27, 77-125. 
Fujita, T. T. (1970): Estimate of Areal Probability of Tornadoes from Inflationary 

Reporting of their Frequencies. SMRP Research Paper No. 89, University of 

Oiicago, 23 pp. 
Fujita, T. T. , D. L. Bradbury and C. F, Van Thullenar (1970): Palm Sunday Tornadoes 

of April 11, 1965. Monthly Weather 'ue, 98, 29-69. 
Hoecker, W. H. , Jr. (1960a): The Dimensional and Rotational CJiaracteristics of the 

Tornadoes and their Cloud System. Research Paper No. 41, U. S. Weather 

Bureau, 53-113. 



40 



Hoecker, W. H. , Jr. , (1960b): Wind Speed and Air Flow Patterns in the Dallas Tornado 

of April 2, 1957. Monthly Weather Revue, 88, 167-180. 
Ishizaki, H. , et al. (1961): Distribution of Structural Wind Damage Caused by the 

Ise-wan Typhoon (in Japanese). Annual Report No. 4, Disaster Prevention Research 

Institute, Kyoto University, Kyoto, Japan, 95-104. 
Japanese Storm Data 1950-59: Kishoyoran (in Japanese). Japan Meteorological Agency, 

Tokyo, Japan. 
List, R. J. (1958): Beaufort Wind Scale, Table 36, Smithsonian Meteorological Tables . 

Smithsonian Institution, Washington, D. C. 
Melaragno, M. G. (1%8): Tornado Forces and their Effects on Buildings. Kansas 

State University, 51 pp. 
Pautz, E. M. (1969): Severe Local Storm Occurrences, 1955-1967. ESSA Tech. 

Memorandum WBTM FSST 12, Office of Meteorological Operation, Silver Springs, 

Maryland. 
Penn, S. , et al. (1%5): The Squall Line and Massachusetts Tornadoes of June 9, 1953. 

Bull. Amer. Meteor. Soc. , 36, 109-122. 
Riehl, H. (1954): Tropical Meteorology. McGraw-Hill Book Co. Inc. , New York, 392 pp 
Segner, E. P. , Jr. , (1960): Estimates of Minimum Wind Forces Causing Structural 

Damage. Research Paper No. 41, U.S. Weather Bureau, 169-175. 
Simpson, R. H. (1970): Personal Communication on Celia Damage. 
St:ats, W. F. and C. M. Turrentine (1956): Some Observations and Radar Pictures of 

Blackwell and Udall Tornadoes of May 25, 1955. Bull. Amer. Meteor. Soc. , 37, 

495-505. 
Tecson, Jaime J. (1971): Characterization of 1965 Tornadoes by their Area and Intensity. 

SMRP Research Paper No. 94, University of Oiicago. 
Thom, H. C. S. (1963): Tornado Probabilities. Mon. Wea. Rev., 91, 730-736. 
Van Tassel, E. L. (1955): The North Platte Valley Tornado Outbreak of June 27, 1955. 

Mon. Wea. Rev. , 83, 239-254. 
Wolford, V. L. (1960): Tornado Occurrences in the United States. Tech. Paper No. 20, 

U. S. Weather Bureau, Washington, D. C. , 71 pp. 



41 



SUBJECT INDEX 



Beaufort Force, 4 

Blackwell Tornado, 16 

Dallas Tornado, 12, 13, 16 

Doppler Winds, 31 

Explosive Pressure, 10, 12 

Fargo Tornado, 3, 14, 15, 16 

Fastest Mile Wind, 5, 24, 29 

Fastest 1/4-Mile Wind, 5, 24, 25, 26 

Fastest 1-Min Wind, 23, 29 

Fastest 10-MinWind, 23, 29 

Fujita Scale 

Application to Hurricanes , 22, 23, 
32, 33, 34, 35, 36 

Application to Tornadoes, 13, 14, 
15, 16, 20, 21 

Conversion Function, 28, 29 

Conversion Table, 30 

Damage Photographs, 11 

Damage Specifications, 8, 9 

Definition, 5 

Equation for Wind Speed, 4 

Speed Table, 6 
Gustiness Factor, 24, 25, 26 
Gustiness Period, 25, 26, 28 
Hurricane Camille, 23, 35 
Hurricane Celia, 32, 33 
Hurricane Gladys, 31 
Ise-wan Typhoon, 25, 33, 34 



Japanese Tornadoes, 19, 21, 22 
Lubbock Tornado, 3 
Mach Number, 4 

Mean Wind, 25, 26, 27, 28, 29, 31, 32 
Palm Sunday Tornadoes, 3, 13, 17, 20, 27 
Peak Gust, 24, 25, 26, 29, 30, 31, 32, 33 
Pressure-Deficit Coefficient, 36, 37 
Rankine-Vortex, 37 
RECONWind, 22, 29, 31, 32 
Scottsbluff Tornadoes, 16 



42